Mmm, I never bothered to listen or measure differences between +/- 15V and +/- 17.5V but I feel the latter is too close to the maximum ratings and a dying opamp causes nasty effects. It is not worth the risk I think. I just make sure they have enough headroom.
No direct comparison but some TDA types sound quite good like TDA7297. LM3875 also sounds good.
BTW I think you will have very good results even when you'll use 7815/7915 (by Onsemi) for the opamp. Can be done quite compact too which is advisable. With CRC or CLC at the inputs (coils instead of resistors for R1/R2).
No direct comparison but some TDA types sound quite good like TDA7297. LM3875 also sounds good.
BTW I think you will have very good results even when you'll use 7815/7915 (by Onsemi) for the opamp. Can be done quite compact too which is advisable. With CRC or CLC at the inputs (coils instead of resistors for R1/R2).
Last edited:
The current drawn on the power amplifier modulates the voltage of the rails which are supplying the zeners etc. The LT3045 demands ~300mV of "overhead" -- so you are right on the cusp.
So, if we are tapping into the main power supply (sharing the same secondary transformer winding, as in the schematic) to power the single opamp, directly or via a good regulator, there is no interaction happening (micro power sagging, crosstalk, modulation, whatever) to deteriorate the sound quality?
In other words, there is no need to have a separate transformer or separate secondary to power the pre amp section of the amplifier?
In other words, there is no need to have a separate transformer or separate secondary to power the pre amp section of the amplifier?
When regulators are used:yes.
One (1) opamp is a very light load which makes it easier. You could go the extreme way but it will be a way of diminishing returns.
One (1) opamp is a very light load which makes it easier. You could go the extreme way but it will be a way of diminishing returns.
What about power supply output impedance? I am looking at those 1k resistors to drop from 24v to 15v. Is that a good way to do that? I read that the Zout must be low, in a good power supply. True? How can it be low if we have the 1k dropping resistors on the +-15v rails?
Because of the low impedance of the Zener diodes. Not the ideal situation as you’ve found out.
There are other ways to have a low Zout 🙂
There are other ways to have a low Zout 🙂
Take a look at the high voltage example of the LT3080 -- the input output voltage is limited by a Zener diode, a MOSFET does the heavy lifting. You should be able to do this with the LT3045, using Rset of 150k
With the example opamps (high PSRR) and only a few mA load current least complex is 30V differential voltage capable regulators isn’t it!?
Jean-Paul,
Yes, but here is what puzzles me:
With the same OPA1642 being used in both situations, only difference is PS.
SilentSwitcher +-15v, powered by 5v usb charger
LT3045/LT3094 type regulator, +-14.5v tapped off the amp's +-15v rails.
https://www.ebay.com/itm/283667518088
Shouldn't both ways sound equally good? They don't. Unfortunately, "SilentSwitcher +-15v, powered by 5v usb charger" sound way better.
Why? can someone hint at the causes?
Yes, but here is what puzzles me:
With the same OPA1642 being used in both situations, only difference is PS.
SilentSwitcher +-15v, powered by 5v usb charger
LT3045/LT3094 type regulator, +-14.5v tapped off the amp's +-15v rails.
https://www.ebay.com/itm/283667518088
Shouldn't both ways sound equally good? They don't. Unfortunately, "SilentSwitcher +-15v, powered by 5v usb charger" sound way better.
Why? can someone hint at the causes?
Wrong layout/design of the Ebay one!?!? Non optimal grounding in that one? Could be a lot of things, not easy to deduct all from pictures.
Maybe the 0.5V is not enough with this design? Wiring too long picking up other stuff? X5R or even worse (piezosensitive) ceramic caps?! Lack of larger value electrolytic caps at both inputs and outputs?
The Silentswitcher was designed with care and for audio. The Ebay one to be cheap 🙂
Maybe the 0.5V is not enough with this design? Wiring too long picking up other stuff? X5R or even worse (piezosensitive) ceramic caps?! Lack of larger value electrolytic caps at both inputs and outputs?
The Silentswitcher was designed with care and for audio. The Ebay one to be cheap 🙂
Last edited:
I honestly dont know how it sounds better than one of the best ic solutions on the market 😁 Any low esr cap on the output of regs?
it's exactly what is in the ebay listing. I did not add any caps on the outputs. Should I?
Usually low esr caps on output of series regulators cause ringing. Hence why i asked.
Yes, add big cap (2200uf) in parallel with 0.1uf wima on input, and 100uf-330uf cap on the output (make sure it isn't low esr).
Now looking at that ebay regulator board, there is no way to deduct what they have on there. As jean paul said, many of those components can cause issues, as well as the layout.
Yes, add big cap (2200uf) in parallel with 0.1uf wima on input, and 100uf-330uf cap on the output (make sure it isn't low esr).
Now looking at that ebay regulator board, there is no way to deduct what they have on there. As jean paul said, many of those components can cause issues, as well as the layout.
LT3045 & LT3094 require low ESR caps on output. No benefit in using large electrolytics there. The layout on that ebay board is bad. The critical Cset capacitor (C4,C11) seems to be at best X5R maybe even worse.
Jan's SilentSwitcher uses TPS7A47 & TPS7A33 which are both very good regulators as well. Having implemented "SilentSwitchers" with both TPS7As and LT3045/LT3094 the difference powering opamps is IME minimal. My guess is that noise in the power amplifier voltage rails and ground creeps through even with regulators so feeding the ebay board with an external power supply could improve the sound.
Jan's SilentSwitcher uses TPS7A47 & TPS7A33 which are both very good regulators as well. Having implemented "SilentSwitchers" with both TPS7As and LT3045/LT3094 the difference powering opamps is IME minimal. My guess is that noise in the power amplifier voltage rails and ground creeps through even with regulators so feeding the ebay board with an external power supply could improve the sound.
The silent switcher designed by Jan has all the hallmarks of a great regulator, high PSRR, low noise and low output impedance.
Try this:
Attachments
Last edited:
zeners work fine for one opamp, not much current.
it is more how much ripple they have to deal with.
main supply caps only 4700u
ripple is what it is, but only at higher currents.
plenty people get wild with capacitor banks
but reality 6800u or 8200u would be more than enough.
and the 15 volt shunt regulators are 22u so just pretty standard
100u would be fine.
no soft start, but this is a case were simple thermistors make people
freak out. but simple thermistor soft start and 8200u caps would wake up the supply.
pretty much doubt regulator changed whole sound stage.
it is more how much ripple they have to deal with.
main supply caps only 4700u
ripple is what it is, but only at higher currents.
plenty people get wild with capacitor banks
but reality 6800u or 8200u would be more than enough.
and the 15 volt shunt regulators are 22u so just pretty standard
100u would be fine.
no soft start, but this is a case were simple thermistors make people
freak out. but simple thermistor soft start and 8200u caps would wake up the supply.
pretty much doubt regulator changed whole sound stage.
Yup that's art least 2 contradictions in 1 post.
noviygera, you could just try the 1 $ 7815/7915 experiment. They don't need heatsinks and may be mounted together on a plastic sheet between them with 1 bolt/nut. Only 6 connections and you'll know. In all cases where I did such the situation improved compared to just Zener diodes. In such a small amplifier the risk is that an external PSU will be larger than the amplifier when going extreme. Also one would not want to have a possible scenario with 2 PSUs that can have a different start up sequence.
Normally one rarely does use 78xx/79xx anymore but in this case it would surprise me if the difference is noticeable when feeding 1 opamp and maximum 10 mA. Certainly with a quality opamp that has high PSRR. 78xx/79xx are less sensitive as well, they just work. The TO92 L versions look to be the better ones for your application but please don't. The regs will be where they are supposed to be and grounding will then also be less of an issue.
And please: Volt, mA, kHz, W, kOhm, µF etc. otherwise we'll look like cavemen with smart phones.
noviygera, you could just try the 1 $ 7815/7915 experiment. They don't need heatsinks and may be mounted together on a plastic sheet between them with 1 bolt/nut. Only 6 connections and you'll know. In all cases where I did such the situation improved compared to just Zener diodes. In such a small amplifier the risk is that an external PSU will be larger than the amplifier when going extreme. Also one would not want to have a possible scenario with 2 PSUs that can have a different start up sequence.
Normally one rarely does use 78xx/79xx anymore but in this case it would surprise me if the difference is noticeable when feeding 1 opamp and maximum 10 mA. Certainly with a quality opamp that has high PSRR. 78xx/79xx are less sensitive as well, they just work. The TO92 L versions look to be the better ones for your application but please don't. The regs will be where they are supposed to be and grounding will then also be less of an issue.
And please: Volt, mA, kHz, W, kOhm, µF etc. otherwise we'll look like cavemen with smart phones.
Last edited:
Here's an inexpensive option, though slightly larger than the silent switcher...
https://www.ebay.com/itm/185583381573?
I've used them on a couple projects. Not sure how much ripple/noise is on the outputs, but I don't get overly concerned with such things unless it causes me obvious problems.
https://www.ebay.com/itm/185583381573?
I've used them on a couple projects. Not sure how much ripple/noise is on the outputs, but I don't get overly concerned with such things unless it causes me obvious problems.
